Videos may be analyzed based on their visual content to identify an activity being performed during video capture. Analyzing visual content may involve techniques that are computationally expensive.